The generative AI model cuts frame rates by nearly 60% in testing while enhancing visual detail.
Nvidia launched DLSS 5 at GTC 2026 on 18 September, introducing generative artificial intelligence to real-time rendering in video games. The technology uses 3D Guided Neural Rendering during the final graphics processing stage to touch up and recreate lighting, shadows, and objects across different materials and surfaces.
The feature debuted commercially in the basketball simulator NBA 2K27, developed by Visual Concepts and published by 2K. The tool is available for GeForce RTX 50 Series graphics cards and laptops, as well as Nvidia's GeForce NOW cloud gaming service. Studios including Capcom, Bethesda, and Ubisoft have backed the technology, with adaptations confirmed for games such as Resident Evil: Requiem, Starfield, and Hogwarts Legacy.
Unlike previous iterations focused on increasing frames per second, the new model prioritises visual detail at the expense of computational performance. Testing conducted by TechSpot revealed that enabling DLSS 5 caused frame rates to drop from 192 to 76 frames per second, representing a decrease of nearly 60%.
Independent analysis by specialised outlet Xataka using an RTX 5090 card showed that visual improvements were subtle during live gameplay, primarily visible during slow-motion replays in sweat detail, softer shadows, and crowd lighting. Edward Liu, director of applied deep learning research at Nvidia, noted that in 4K testing the neural rendering cut frame rates by up to 60%, requiring graphics processors to rely on artificial frame generation to maintain fluidity.
Newsletter
Markets in your inbox, weekly
LATAM-focused analysis, investing ideas, and the week in finance.
Keep reading